Shrink down lookup tables

Author: sfackler

codegen/src/main.rs

@@ -4,77 +4,85 @@ use regex::Regex;
 
 use std::fs::File;
 use std::io::prelude::*;
-use std::io::{BufRead, BufWriter};
+use std::io::BufWriter;
 
 // Generate character mapping tables directly from the specification.
 fn main() {
-    // Input from the RFC.
-    let reader = include_bytes!("rfc3454.txt");
-
-    // Output to a Rust source file.
     let out_file = File::create("../src/rfc3454.rs").unwrap();
     let mut writer = BufWriter::new(out_file);
 
-    // Generate tables.
-    include_table(&mut writer, &mut &reader[..], "A.1");
-    include_table(&mut writer, &mut &reader[..], "B.2");
+    write!(writer, "// AUTOGENERATED CODE - DO NOT EDIT\n\n").unwrap();
+
+    lookup_table("A.1", &mut writer);
+    mapping_table("B.2", &mut writer);
 }
 
-// Generate code for the named mapping table.
-fn include_table<R: BufRead, W: Write>(writer: &mut W, reader: &mut R, tablename: &str) {
-    // Scan to start of table.
-    loop {
-        let mut line = String::new();
-        reader.read_line(&mut line).unwrap();
-        if line.contains("Start Table") && line.contains(tablename) {
-            break;
+fn lookup_table<W: Write>(table: &str, writer: &mut W) {
+    write!(
+        writer,
+        "pub const {}: &'static [(char, char)] = &[\n",
+        table.replace(".", "_")
+    ).unwrap();
+
+    let regex = Regex::new("^([0-9A-F]+)(?:-([0-9A-F]+))?$").unwrap();
+    table_lines(table, |line| {
+        let captures = regex.captures(line).unwrap();
+        let start = captures.get(1).unwrap().as_str();
+        let end = captures.get(2).map_or(start, |c| c.as_str());
+        write!(writer, "    ('\\u{{{}}}', '\\u{{{}}}'),\n", start, end).unwrap();
+    });
+
+    write!(writer, "];\n\n").unwrap();
+}
+
+fn mapping_table<W: Write>(table: &str, writer: &mut W) {
+    write!(
+        writer,
+        "pub const {}: &'static [(char, &'static str)] = &[\n",
+        table.replace(".", "_"),
+    ).unwrap();
+
+    let regex = Regex::new(
+        "^([0-9A-F]+); ([0-9A-F]+)(?: ([0-9A-F]+))?(?: ([0-9A-F]+))?(?: ([0-9A-F]+))?;",
+    ).unwrap();
+    table_lines(table, |line| {
+        let captures = regex.captures(line).unwrap();
+        let mut it = captures.iter().filter_map(|i| i);
+        it.next(); // skip whole match
+
+        let input = it.next().unwrap().as_str();
+        write!(writer, "    ('\\u{{{}}}', \"", input).unwrap();
+
+        for output in it {
+            write!(writer, "\\u{{{}}}", output.as_str()).unwrap();
         }
-    }
 
-    // Output table declaration.
-    write!(writer, "pub const {}: &[(char, char, &str)] = &[\n", tablename.replace(".", "_")).unwrap();
+        write!(writer, "\"),\n").unwrap();
+    });
+
+    write!(writer, "];\n\n").unwrap();
+}
 
-    // For each line:
-    let target_re = Regex::new(r"([0-9A-F]+)(-([0-9A-F]+))?(; ([0-9A-F]+)( ([0-9A-F]+))?( ([0-9A-F]+))?( ([0-9A-F]+))?;)?").unwrap();
-    loop {
-        let mut line = String::new();
-        reader.read_line(&mut line).unwrap();
+fn table_lines<F>(table: &str, mut f: F)
+where
+    F: FnMut(&str),
+{
+    let mut lines = include_str!("rfc3454.txt").split('\n');
 
-        // Done when reach the end of the table.
+    // fast forward to the start of the table
+    lines.find(|line| line.contains("Start Table") && line.contains(table));
+
+    for line in lines {
+        let line = line.trim();
         if line.contains("End Table") {
             break;
         }
 
-        // Skip RFC metadata.
-        if line.contains("Hoffman & Blanchet") || line.contains("RFC 3454") {
+        // Skip page headers/footers
+        if line.is_empty() || line.contains("Hoffman & Blanchet") || line.contains("RFC 3454") {
             continue;
         }
 
-        // Generate an entry for each data line.
-        if let Some(captures) = target_re.captures(&line) {
-            // start char
-            let start = captures.get(1).unwrap().as_str();
-
-            // end char (inclusive)
-            let end = captures.get(3).map_or(start, |m| m.as_str());
-
-            // 0-4 character replacement string
-            let mut replace = String::new();
-            for &i in [5, 7, 9, 11].iter() {
-                match captures.get(i) {
-                    None => break,
-                    Some(c) => {
-                        replace.push_str("\\u{");
-                        replace.push_str(c.as_str());
-                        replace.push_str("}");
-                    }
-                }
-            }
-
-            write!(writer, "    ('\\u{{{}}}', '\\u{{{}}}', \"{}\"),\n", start, end, replace).unwrap()
-        }
+        f(line);
     }
-
-    // End table definition.
-    write!(writer, "];\n\n").unwrap();
 }